The environment contains far more information than the brain can process at once. To cope with such information overload, humans need to selectively attend to relevant information and prioritize its processing. In many situations, humans need to select arbitrary features and objects in the scene and maintain attention on the selected information. It is often assumed that an attentional priority signal encodes the current focus of attention and its deployment. However, how the brain computes and maintains attentional priority for features and objects is not known. The long-term goal is to understand how the brain selects different types of information and how selection shapes perception to serve goal-oriented behavior. The objective of this proposal is to delineate the cortical circuitry representing attentional priority for features and objects using functional magnetic resonance imaging (fMRI). Based on recent data obtained in our laboratory, we hypothesize that the dorsal frontoparietal network represents different types of selected information with distinct neural populations, forming a multiplexed representation of attentional priority. In this proposal, we wil test this hypothesis by pursuing four specific aims. First, we will determine the neural representation of attentional priority for visual objects. Second, we will seek to establish a quantitative link between priority signals and task performance. Third, we will determine the relationship between attentional priority signals for features and objects and those for spatial locations. Fourth, we will evaluate the degree of categorical representation of attentional priorit, which is essential for flexible deployment of attention. The proposed research is expected to significantly advance our understanding of how the brain selects and maintains non- spatial information, thus filling in a critical gap in the current scientific knowledge. A deeper understanding of how the brain selects features and objects will provide important constraints for models of attention and can potentially transform our understanding of visual information processing and cognitive control. The proposed research is innovative both in terms of conceptual and methodological advances. Conceptually, the research will test the novel hypothesis that the dorsal frontoparietal network represents attention priority for non-spatial dimensions, challenging the prevailing view that these cortical areas mainly represent spatial information. Methodologically, the application of cutting-edge machine learning and data mining techniques (pattern classification, similarity and clustering analysis) represents a novel approach that more fully exploits the complexity and richness of fMRI data than conventional methods. Finally, the proposed research can make connections to other fields such as category learning and decision making, and suggest interesting future directions to examine common neural processes underlying these cognitive functions.